NO2 and NO Adsorption Properties of KOH ... - ACS Publications

NO2 and NO on γ-alumina before and after KOH treatments has been evaluated. ... stream significantly enhances the adsorption properties of the KOH-tr...
0 downloads 0 Views 133KB Size
Download PDF
Ind. Eng. Chem. Res. 1998, 37, 3375-3381

3375

MATERIALS AND INTERFACES NO2 and NO Adsorption Properties of KOH-Treated γ-Alumina Maxwell R. Lee,† Eric R. Allen,† John T. Wolan,‡ and Gar B. Hoflund*,‡ Departments of Environmental Engineering Sciences and Chemical Engineering, University of Florida, Gainesville, Florida 32611

A method to control nitrogen oxide (NOx) emissions from combustion sources by adsorption of NO2 and NO on γ-alumina before and after KOH treatments has been evaluated. Compared to previously studied sorbents consisting of magnesium-oxide-coated vermiculite, untreated γ-alumina exhibits a 6-fold increase in activity in tubular flow system tests. XPS analyses of the adsorbent surfaces before and after exposure to NOx indicate that potassium influences the NOx-sorption process. Subsequent treatment of γ-alumina with KOH by impregnation or precipitation improves the adsorptive properties of γ-alumina toward both NO2 and NO, with the precipitated samples performing better than impregnated samples. This research confirms previous findings that sorption of 3 mol of NO2 on γ-alumina results in the catalytic formation of 1 mol of NO. However, treatment with KOH delays and reduces the formation of NO while increasing 5-fold the amount of NO2 adsorbed. Formation of nitrate and nitrite species is observed by XPS analysis of KOH-precipitated γ-alumina exposed to NOx. A 40% loss of BET surface area occurs due to KOH precipitation on γ-alumnia followed by a further 56% loss in surface area after saturation with NOx. The addition of water vapor (3 vol %) to the feed gas stream significantly enhances the adsorption properties of the KOH-treated γ-alumina. Washing the γ-alumina pellets exposed to NOx with water essentially removes all of the potassium nitrates and nitrites formed. This harmless solution can be disposed of safely or used as fertilizer, and the pellets can be regenerated. Introduction The emissions of nitrogen oxides (NOx) from combustion sources are of concern because they are precursors of photochemical ozone, contribute to acid rain, and pose a threat to public health as contributors to respiratory ailments. NOx emitted from combustion processes predominantly consists of nitric oxide and a smaller amount of nitrogen dioxide. These are produced by high-temperature oxidation of gaseous molecular nitrogen and nitrogen contained in fuels. Numerous technologies have been developed to control NOx emissions by modification of the combustion process conditions and/or postcombustion NOx removal. Postcombustion NOx control technologies have distinct disadvantages that limit the universal application of any particular method. A commonly sought goal of postcombustion NOx control systems is the reduction of NOx to harmless species through selective catalytic reduction or selective noncatalytic reduction methods using a reducing gas. Applications of catalytic NOx reduction methods are hindered, however, by the requirement of steady-state combustion conditions to maintain an appropriate reaction temperature and composition. Incomplete or disproportionate mixing of reducing gas and flue gas can result in the emission of the reducing gas or enhanced NOx formation. Other considerations for catalytic re* Corresponding author. † Department of Environmental Engineering Sciences. ‡ Department of Chemical Engineering.

duction include catalyst deactivation and the cost of installation and maintenance. Noncatalytic reduction typically requires a higher and narrower operational temperature range than catalytic reduction and is also susceptible to disadvantages similar to those for catalytic reduction. Another approach is to oxidize NO to NO2 and then use wet sorption to remove the NO2. The provision of oxidizing material and handling and disposal of wet, spent material contribute to the cost of wet sorption methods. Hybrid systems incorporating multiple techniques to collectively reduce NOx emissions have also been evaluated.1 Increasingly stringent regulations on NOx emissions from mobile and stationary sources have led to the development of novel control techniques as discussed by Markussen and Livengood.2 Control of NOx by chemisorption on solid surfaces has been attempted in laboratory- and pilot-scale studies.3-7 Chemisorption methods have several common advantages including ease of material handling, simplicity of system design, and low maintenance requirements. Testing of adsorbent materials has shown that they provide for NOx removal over a greater range of operating conditions without the possibility of adverse emissions of reducing agents or the necessity of compositional monitoring controls compared to selective reduction methods.7 Clearly, an inexpensive, readily available sorbent with efficient sorptive characteristics would be useful in NOx control applications for numerous types of emission sources.

S0888-5885(98)00104-3 CCC: $15.00 © 1998 American Chemical Society Published on Web 07/17/1998

3376 Ind. Eng. Chem. Res., Vol. 37, No. 8, 1998

The United States Air Force has promoted a number of studies to control NOx emissions from jet engine test cells (JETC).4-14 Conventional NOx control technologies are not applicable in this case due to large variations in operating conditions required in these test cells.15 The most promising material developed and tested in these studies was an adsorbent consisting of vermiculite treated with magnesium oxide (MgO/v).7 MgO/v was pilot-scale tested and found to perform with varied efficiency (30-70%) for removal of NOx over the range of operating conditions and allowable pressure drop in JETC testing.6 Sorbent Technologies Corp. (Twinsburg, OH) (Sorbtech) has patented this solid NOx adsorbent16 and has performed a variety of tests for NOx removal from various flue gas sources.4-7 MgO/v typically consists of 50% (by mass) vermiculite, which primarily acts as a high-surface-area support for the NOx-reactive MgO.10 Vermiculite, which is a naturally occurring brittle mineral that expands when heated under moist conditions,17 has a poor compressive strength, resulting in pellets that are easily degraded with minimal handling. The degraded sorbent may be susceptible to emission in the flue gas as fine particulate material. Another disadvantage is that the prepared sorbent has a fairly low surface area of about 40 m2/ g.10 Thus, although MgO/v has the potential to limit NOx emissions over a wide range of combustion conditions, it suffers from several disadvantages. It is brittle, has a naturally variable composition, must function below 250 °C, and has limited capacity for NOx removal. The limitations of MgO/v as a sorbent for NOx have led to a search for other possible sorbents. Desirable characteristics include high surface area, low cost, wide availability, structural integrity, and good sorptive properties for NOx. γ-Alumina is a potential candidate based on the results of previous chemisorption studies.18,19 Similar sorbent applications of alumina for removing reactive gas stream constituents, such as sulfur dioxide, also have been reported.20-22 In recent laboratory- and pilot-scale studies,23-27 it was found that Na2CO3-treated γ-alumina provides up to 90% removal of NOx and SO2 from flue gas under the conditions used. In this present study γ-alumina pellets before and after different treatments with KOH have been studied for adsorption of NOx, and improved surface treatments have been developed based on surface characterization results obtained using X-ray photoelectron spectroscopy (XPS). Experimental Methods Two experimental systems were used to evaluate the performance of NOx sorbents. An isothermal tubular flow system was used to expose sorbent samples to mixtures of NO and NO2 in nitrogen. Certified-gascylinder mixtures of NO (4980 ppm) in N2 and NO2 (5028 ppm) in N2 (BOC Gases, Riverton, NJ) were diluted to desired concentrations (500 ppm NO2 and 50 ppm NO in N2) using calibrated rotameters. The total gas flow was measured with a digital mass flowmeter (Tylan model FC-280, Rancho Dominguez, CA). All flow measuring devices were calibrated in-line using either a primary flow standard (Gilian Instrument Corp., model 2373-B gilibrator, Wayne, NJ) or a calibrated wettest meter (Precision Scientific model 63126, Bellwood, IL). The gas mixture flowed through a switching valve that directed it either to a bypass line or to the sorbent beds.

Gas mixtures were split equally using a manifold leading to three parallel beds. The 16-in.-long parallel tubes containing the 4.5-in.-long sorbent beds (0.65-in. i.d.) were mounted vertically and symmetrically inside a temperature-controlled furnace (Thermolyne model F79325, Dubuque, IA). Both the connecting tubing and sorbent-bed tubes were made of 316 stainless steel. Bed temperatures during tests were monitored with a type-K thermocouple (Omega Engineering Inc. model KQSS116G, Stamford, CT) inserted into each bed through Ultra-Torr fittings. Preliminary tests showed that minimal temperature gradients were obtained for the experimental flow rates and temperatures used when the beds were placed in the middle section of the tubes. The bed temperature profiles were significantly altered by changing the total gas flow rate. Thus, a 1-min nitrogen flush of the lines was performed prior to each test to stabilize the temperature profile, allow for the fine adjustment of bed outlet gas flows, and remove stagnant bed gas. The sorbent material was supported on a 316 stainless steel wire mesh support. Gas flowing out of the beds could be switched to flow-measuring devices or to gas-sampling solenoid valves for subsequent NOx analysis. Chemiluminescent detection of bed inlet and outlet NO and NO2 concentrations was performed using a NOx analyzer (Thermo Environmental Instruments, Inc., model 42 H, Franklin, MA). The NOx analyzer was calibrated daily using certified gas cylinders of NO (797 ppm in N2) and NO2 (741 ppm in N2) (BOC Gases, Riverton, NJ). Pre- and postrun checks of bed inlet NO and NO2 concentrations were performed to ensure that variations less than (5% occurred during the experiments. The possibility of interaction of NOx with the system materials was tested by comparing the NOx-uptake levels with that obtained by feeding the certified NOx gas mixture directly to the analyzer. This showed that internal NOx losses were consistently less than 1%. Experiments were performed to test whether NOx sorption under the conditions used was limited by mass transfer. This occurs if the sorption process varies in tests in which bed residence time remains constant while flow rate is varied. Tests of this type demonstrated that mass-transfer effects were negligible. For some runs water vapor (3 vol %) was added to the gas mixture entering the tubular system to evaluate its effect on NOx adsorption. In all cases the total gas flow rate was 1.5 L/min at the bed conditions. XPS was utilized to characterize the compositions of the sorbent surfaces before and after addition of the KOH and exposure to NOx. Prior to XPS analysis, alumina pellets were crushed to expose the inner surface, placed in an aluminum cup, pressed into a new pellet, mounted on a stainless steel sample holder, and inserted into an ultrahigh-vacuum (UHV) chamber (base pressure 10-10 Torr). This procedure exposes the pore walls for analysis. XPS was performed using a double-pass cylindrical mirror analyzer (CMA) (PerkinElmer PHI model 25-270AR, Eden Prairie, MN). Survey spectra were taken in the retarding mode with a pass energy of 50 eV, and high-resolution XPS spectra were taken with a pass energy of 25 eV using Mg KR X-rays (Perkin-Elmer PHI model 04-51 X-ray source). Data collection was accomplished using a computerinterfaced, digital pulse-counting circuit28 followed by smoothing using digital filtering techniques.29 Since

Ind. Eng. Chem. Res., Vol. 37, No. 8, 1998 3377

XPS was performed using a flood source, the XPS data represent an average over the area analyzed (∼4-mm diameter). Sorbent Preparation MgO/v was prepared by the following technique.10 One-kilogram batches were prepared using a 45:55 mass ratio of MgO (Elastomag 170 m2/g, Akrochem Corp., Akron, OH) and vermiculite (coarse grade, Stronglite Product Corp., Pine Buff, AR). Each batch was handmixed with a 4:1 mass ratio of deionized water/MgOvermiculite. Batches were baked at 550 °C for 30 min in a muffle furnace and stored in airtight plastic containers. This preparation method is described in more detail in the patent description16 held by Sorbtech. Sorption studies of the effects of aging on the MgO/v material stored in airtight containers showed that the stored sorbent performs well without deterioration over a minimum period of 2 years. Storage in ambient air for periods of several weeks also does not alter the NOx sorption characteristics. Cylindrical pellets (1/8-in. o.d. × 1/8 in.) of γ-alumina were obtained commercially (Material AL-3438 T, Englehard Corp., Elyria, OH) with a nominal surface area of 170 m2/g. Pellets were dried at 150 °C for 24 h prior to sorbent preparation or storage in sealed containers. Reagent-grade KOH (Fischer Scientific, Pittsburgh, PA) was used to make aqueous solutions for treating the alumina pellets. Solution concentrations varied from 1.0 to 0.04 M. Two types of treatment techniques were used: impregnation and precipitation. Impregnated pellets were prepared by flowing the KOH solutions through beds of alumina pellets supported on Whatman 41 filter paper for about 2 min. After impregnation and drying, the sample weight was increased by about 2%, and this was increased by exposure to larger volumes of the KOH solution. Both impregnated and precipitated samples were dried in air at 150 °C for 24 h. Precipitated samples were prepared by mixing pellets with the KOH solution and evaporating the solution to dryness. The weight gain of precipitated samples used for NOx exposure was 24%. Precipitated samples prepared using KOH solutions of concentrations greater than 1 M exhibit physical deterioration, and saturated KOH solutions dissolve γ-alumina to form an amorphous sludge. The samples were stored in airtight plastic containers prior to use. NOx sorption tests on precipitated alumina pellets having similar KOH loadings but prepared using different solution concentrations demonstrated that they have similar NOx removal characteristics. The nature of the containers (glass or ceramic) used to prepare treated samples does not alter NOx sorption behavior. Bulk densities for MgO/v (sieve size, Tyler no. 8), untreated γ-alumina, and KOHprecipitated alumina (24% mass loading) are 0.13, 0.70, and 0.88 g/cm3, respectively. Results and Discussion Initial NOx sorption experiments were performed on MgO/v and untreated γ-alumina. Equal-volume beds (25 cm3) of these sorbents were exposed to a gas stream containing 500 ppm NO2 and 50 ppm NO in nitrogen at a temperature of 200 °C. The experimental mole ratio of NO/NO2 used in these experiments of 50 ppm to 500 ppm (1:10) is not typical of combustion-source NOx, in which the ratio is typically near 20:1. However,

Figure 1. Bed outlet NO and NO2 concentrations as a function of time for MgO/v and γ-alumina. Test conditions: 200 °C, 1-s residence time, 25-cm3 bed volume, 500 ppm NO2 and 50 ppm NO in nitrogen.

in the proposed design of a suitable NOx sorbent system, an oxidizing procedure would precede the NOx sorbent to transform NO into NO2. Assuming incomplete preoxidation of NO to NO2, the experimental NO/NO2 ratio of 1:10 was chosen for initial experimentation. The NO2 and NO bed outlet concentrations obtained as a function of NOx-exposure time are shown in Figure 1. Following an initial decrease to zero, the NO2 concentration in the stream leaving the MgO/v bed initially increases rapidly and then more slowly, attaining a value of about 430 ppm after 300 min. After initial fluctuations the NO concentration increases rapidly to a maximum of 190 ppm at 20 min and then decreases slowly to about 50 ppm, which is the NO feed concentration. The fact that the NO concentration rises above 50 ppm indicates that NO2 is being catalytically converted into NO in the bed. The untreated γ-alumina bed initially removes all NO2 from the gas stream, and then the NO2 concentration increases gradually to a value of 100 ppm at 300 min. Initially, the γ-alumina removes all of the inlet NO. Then the NO concentration increases fairly rapidly to a level near 200 ppm at about 100 min, indicating that γ-alumina also catalyzes the formation of NO from NO2. Untreated γ-alumina and MgO/v actively form significant quantities of NO while removing NO2. In addition to providing better NOx sorption than MgO/v, the untreated γ-alumina bed exhibits a 20% lower pressure drop than the MgO/v bed. At saturation (about 400 min for MgO/v and about 1100 min for untreated γ-alumina), the total mole ratio of NO2 removed to NO formed from the gas stream is 3:1 for both MgO/v and untreated γ-alumina. Similar results have been reported for other sorbent materials. Kimm and coworkers9,10 found that NO2 sorption on MgO/v in the absence of oxygen forms NO in the same molar ratio of 3:1 (NO2 removal to NO formation) and proposed a mechanism for NO2 uptake and NO production with the following overall stoichiometry.

3NO2(g) + MgO(s) f Mg(NO3)2(s) + NO(g) (1) Nelli and Rochelle3 found that NO2 adsorbs on various nonalkaline solids including γ-alumina and suggested that sorbed water and NO2 react to produce nitric acid and NO as follows:

3NO2(g) + H2O(l) f 2HNO3(l) + NO(g)

(2)

3378 Ind. Eng. Chem. Res., Vol. 37, No. 8, 1998

Figure 2. XPS survey spectra obtained from (a) fresh and (b) NOxsaturated γ-Al2O3 pressed powder samples. Assignments made pertain to both spectra.

According to Nelli and Rochelle, the production of NO is based on the vapor-liquid equilibrium values of nitrous acid. Boehm30 has explained the formation of nitrate ions as a result of NO2 sorption on alumina as due to a disproportionation of the surface OH-:

[OH-]surf + 3NO2 f [NO3 ]surf + HNO3 + NO

(3)

All of the above equations indicate that 3 mol of NO2 forms 2 mol of nitrate or nitric acid species and 1 mol of NO. The active sites on alumina have been described as basic (hydroxyl and O2- vacancies) and acidic (unsaturated Al3+ ions as Lewis and protonated hydroxyl ions as Bro¨nsted) sites of various strengths and concentrations.31 The hygroscopic natures of alumina and MgO have been well studied and for alumina is known to significantly influence its reactivity.19,31-33 At 50% relative humidity, activated alumina can adsorb significant quantities of water and surface hydroxyl ion species can exist up to temperatures of 650 °C.19 Thus, the reaction of NO2 with alumina at the temperatures used in these experiments (200 °C) may involve adsorbed water or surface OH- species and be consistent with reaction (1), (2), or (3). The reactions of alumina with NO2 and NO have been studied using infrared spectral analysis.34,35 Knozinger34 suggested that NO2 adsorbs to γ-alumina as nitrate or nitrite species. The adsorption of NO on alumina has also been studied. Adsorbed NO produces weakandpoorlydefinedspectralfeaturesonγ-alumina.35-37 In contrast, adsorbed NO2 species produce intense and well-defined spectral features. Thus, Parkyns35 attributed weak NO signals to weakly bound nondissociated species. Based on qualitative comparison, infrared studies confirm the formation of nitrate and nitrite species consistent with the reactions discussed above as do the XPS data presented below. XPS survey spectra obtained from nonexposed and NOx-saturated untreated γ-Al2O3 are shown in parts a and b of Figure 2, respectively. Peak assignments in Figure 2 pertain to both spectra. The spectra include Al 2s and 2p, C 1s, and O 1s and 2s peaks. The C 1s peaks in both spectra are probably due to the presence

Figure 3. Bed outlet NO and NO2 concentrations as a function of time for untreated and KOH-treated γ-alumina. Test conditions: 200 °C, 1-s residence time, 25-cm3 bed volume, 500 ppm NO2 and 50 ppm NO in nitrogen.

of about 2% graphite component from the manufacturing process. The N 1s peak is apparent only after NOx exposure in Figure 2b. The small K 2p peak in Figure 2a is enhanced after exposure to NOx in Figure 2b. Potassium has been identified as a common contaminant in aluminum foils.38 The observation of this significant increase in the K 2p peak after treatment with NOx indicates that there is a strong chemical interaction between surface K and NOx and, therefore, provided justification for examining the effects of pretreating γ-Al2O3 with a potassium compound. Krizek39 evaluated absorption of NO2 by aqueous KOH and found that potassium nitrite is formed. He also found that aqueous KOH more efficiently absorbs NO2 than aqueous NaOH, which has been used for ambient NO2 measurement in past EPA reference methods40 and for flue-gas NO2 measurement.41 Furthermore, Nelli and Rochelle3 suggested that an alkaline solid, such as Ca(OH)2, buffers the pH of calcium silicate. The Ca(OH)2-buffered sorbent maintains a pH high enough to keep HNO2 and HNO3 dissociated, thereby minimizing the production of gaseous NO. Since this previous research has shown that hydroxide compounds can prevent the production of NO and XPS analysis indicates that there is a strong chemical interaction between K and NOx resulting in the enhanced K 2p feature observed in Figure 2b, KOH was chosen for impregnation of γ-alumina. KOH-treated γ-alumina was tested for NOx removal under conditions similar to those used for comparing MgO/v and untreated alumina. The data obtained from untreated γ-alumina, KOH-impregnated γ-alumina, and KOH-precipitated γ-alumina are shown in Figure 3. All three beds yield very low outlet NO2 concentrations for the first 40 min during exposure. After 40 min the outlet NO2 concentrations are the same for KOHimpregnated and -precipitated γ-alumina within experimental error and gradually increase to 50 ppm at 300 min. The KOH-impregnated γ-alumina NO outlet concentration indicates removal of NO for the first 140min period and increases to about 150 ppm at 300 min, which is lower than that of untreated γ-alumina. KOHprecipitated γ-alumina removes nearly all NO over the time period examined. KOH-precipitated γ-alumina exhibits a 90% removal of both NO and NO2 at 300 min, at which time approximately 12 mol of NOx-containing gas has been treated.

Ind. Eng. Chem. Res., Vol. 37, No. 8, 1998 3379 Table 1. NOx Saturation Capacity of Untreated and KOH-Treated γ-Alumina

γ-alumina

NO2 adsorbed per NO formed (mmol/mmol)

treatment mass of KOH (% of sample)

mol NO2 adsorbed/ mol of KOH

nonexposed

NOx saturated

untreated KOH-impregnated KOH-precipitated

1.5/0.54 1.96/0.65 7.8/2.4

0 0.1 g (2) 0.96 g (20)

1.1 0.44

170 163 98

170 154 39

Figure 4. XPS survey spectra obtained from (a) KOH-impregnated and (b) NOx-saturated KOH-impregnated γ-Al2O3 pressed powder samples. Assignments made pertain to both spectra.

XPS survey spectra obtained from the nonexposed KOH-impregnated and NOx-saturated KOH-impregnated γ-Al2O3 are shown in parts a and b of Figure 4, respectively. Peak assignments in Figure 4 pertain to both spectra. The Al 2s and 2p, C 1s, O 1s and 2s, and K 2p peaks are apparent in these spectra. The surface carbon content of this sample is large since the C 1s peak is similar in size to the aluminum peaks and carbon has a low XPS sensitivity. The K 2p peak is enhanced after exposure to NOx as observed in Figure 4b. This is consistent with the XPS data obtained from untreated alumina (Figure 2). The N 1s peak is barely apparent after the NOx exposure. XPS survey spectra obtained from the nonexposed 1 M KOH-precipitated and NOx-saturated KOH-precipitated γ-Al2O3 are shown in parts a and b of Figure 5, respectively. Before exposure the survey spectrum is quite similar to that obtained from the nonexposed KOH-impregnated γ-alumina except that this surface contains less carbon. After NOx exposure very prominent K 2p and 2s and N 1s peaks are apparent, and the C 1s peak height also is increased. High-resolution N 1s spectra obtained from nonexposed KOH-precipitated and KOH-precipitated, NOxsaturated γ-Al2O3 samples are shown in parts a and b of Figure 6A, respectively. The presence of nitrate and nitrite chemical states is clearly observed. The NO3/ NO2 ratios are approximately 1.5:1 and 2.0:1 respectively in a and b, indicating that more nitrate forms during the NOx adsorption. The K 2p feature shown in Figure 6B is significantly increased in size after NOx exposure similar to the N 1s features. The binding energy of the K 2p peaks suggest that it is present as nitrate and/or nitrite before and after the NOx exposure, but it may be present as hydroxide before exposure. As stated above, Nelli and Rochelle3 proposed that a Ca(OH)2-buffered sorbent has a pH high enough to keep

BET surface area (m2/g)

Figure 5. XPS survey spectra obtained from (a) KOH-precipitated and (b) NOx-saturated KOH-precipitated γ-Al2O3 pressed powder samples. Assignments made pertain to both spectra.

Figure 6. XPS (A) N 1s and (B) K 2p spectra obtained from (a) KOH-precipitated and (b) NOx-saturated KOH-precipitated γ-Al2O3 pressed powder samples.

HNO2 and HNO3 dissociated. The decompositions of HNO2 and HNO3 are rate limiting in the mechanism for the overall reaction (2). Even though buffering compounds such as Ca(OH)2 or KOH keep HNO2 and HNO3 dissociated, the experiments of Nelli and Rochelle and the research presented in this paper demonstrate that NO is eventually formed with continued exposure to NO2. Results of saturation testing in the tubular flow system containing γ-alumina, both untreated and treated, are shown in Table 1. The sorbent saturation capacity is defined in this study at the point when the bed NOx outlet concentration does not differ by more than 5% from the inlet concentration. These tests were conducted at the same temperature and NOx concentration used in previous tests. However, to conserve cylinder gas resources, smaller sample masses of 5 g were used. Preliminary tests indicate that the quantity of NO2

3380 Ind. Eng. Chem. Res., Vol. 37, No. 8, 1998

removed per unit sample mass does not vary for different sample masses of untreated alumina saturated with NOx. The amount of NO2 removed per unit amount of added KOH is reduced with increased KOH loading. BET surface area measurements indicate that the surface area is reduced as the KOH loading is increased and that NOx exposure further reduces the surface areas, which may explain the reduction in efficiency. The increased molar volume of nitrate species formed by NOx sorption also has been suggested as a reason for reduced efficiency.10 Since all combustion product streams contain water, the effect of water on the performance of the KOHprecipitated γ-alumina was tested. Addition of 3 vol % water vapor to the feed gas results in a significant improvement in the NOx sorbent properties of this material. The removal of NO is increased by 40% while the removal of NO2 is improved by 20%. In order for these materials to be useful in various applications, it is necessary to be able to regenerate the NOx-exposed γ-alumina so that the γ-alumina can be reprocessed and used repeatedly. Preliminary experiments have demonstrated that the NOx-exposed γ-alumina can be regenerated in two ways. The first is simply washing the NOx-exposed sorbent with water to form a solution of potassium nitrates and nitrites which is environmentally harmless and can be used as a fertilizer. The second way is to heat the NOx-exposed γ-alumina to about 400 °C to form NO and NO2 by decomposition of the surface nitrates and nitrites. The concentrated NO and NO2 could then be converted to nonpolluting species. Summary The NOx adsorption properties of γ-alumina before and after KOH treatment have been examined and compared with those of MgO/v in order to develop improved adsorbents for NOx control. γ-Alumina performs significantly better than MgO/v, and KOH treatments further enhance the performance of γ-alumina. The best adsorbent for NO2 and NO prepared in this study is KOH-precipitated γ-alumina, which exhibits a 5-fold improvement over untreated γ-alumina. XPS studies have been crucial with regard to discovering the importance of the KOH treatment and useful in characterizing the adsorbed NOx species. Acknowledgment Funding for collection of the adsorption data shown in Figures 1 and 3 was provided through a graduate fellowship from the National Science Foundation and USAF (FC8637-96-C-6015). The data collection effort of Jacob Mauldin is also appreciated. Literature Cited (1) U.S. EPA. Alternative Control Techniques Document-NOx Emissions from Utility Boilers; Report No. EPA-453/R-94-023; U.S. EPA: Washington, DC, 1994a. (2) Markussen, J. M.; Livengood, C. D. Alternative Flue Gas Treatment Technologies for Integrated SO2 and NOx Control. Proceedings of the 57th Annual American Power Conference, Chicago, IL, 1995. (3) Nelli, C. H.; Rochelle, G. T. Nitrogen Dioxide Reaction with Alkaline Solids. Ind. Eng. Chem. Res. 1996, 35, 999.

(4) Nelson, B. W.; Nelson, S. G.; Van Stone, D. A. Controlling Combustion-Source Emissions at Air Force Sites with a New Filter Concept; Al/EQ-TR-1994-0006; HQ AFESC: Tyndall AFB, FL, 1994. (5) Nelson, S. G.; Van Stone, D. A.; Little, R. C.; Peterson, R. A. Laboratory Evaluation of a Reactive Baffle Approach to NOx Control; AL/EQ-TR-1993-0017; HQ AFESC: Tyndall AFB, FL, 1993. (6) Nelson, B. W.; Nelson, S. G.; Van Stone, D. A. Development and Demonstration of a New Filter System to Control Emissions During Jet Engine Testing; CEL-TR-92-49; HQ AFESC: Tyndall AFB, FL, 1992. (7) Nelson, B. W.; Nelson, S. G.; Higgins, M. O.; Brandum, P. A. A New Catalyst for NOx Control; ESL-TR-89-11; HQ AFESC: Tyndall AFB, FL, 1989. (8) Lyon, R. K. New Technology for Controlling NOx from Jet Engine Test Cells; ESL-TR-89-15; HQ AFESC: Tyndall AFB, FL, 1991. (9) Kimm, L. T.; Allen, E. R.; Wander, J. D. Control of NOx Emissions from Jet Engine Test Cells. Proceedings of the 88th Annual Air and Waste Management Association Meeting, San Antonio, TX, 1995. (10) Kimm, L. T. Control of Nitrogen Oxide Emissions from Jet Engine Test Cell Exhaust using a Magnesium Oxide-Vermiculite Sorbent. Ph.D. Dissertation, University of Florida, Gainesville, FL, 1995. (11) Petrik, M. A. Use of Metal Oxide Electrocatalysts to Control NOx Emissions from Fixed Sources; ESL-TR-89-29; HQ AFESC: Tyndall AFB, FL, 1991. (12) Kittrell, J. R. High-Temperature NOx Control Process; ESLTR-89-36, HQ AFESC: Tyndall AFB, FL, 1991. (13) Johnson, S. A.; Katz, C. B. Feasibility of Reburning for Controlling NOx Emissions from Air Force Jet Engine Test Cells; ESL-TR-89-33; HQ AFESC: Tyndall AFB, FL, 1989. (14) Berman, E.; Dong, J.; Lichtin, N. N. Photopromoted and Thermal Decomposition of Nitric Oxide by Metal Oxides; ESL-TR91-32; HQ AFESC: Tyndall AFB, FL, 1991. (15) U.S. EPA. NOx Emissions and Their Control from Uninstalled Aircraft Engines in Enclosed Test Cells; Report No. EPA/ 453R-94-008; U.S. EPA: Washington, DC, 1994b. (16) Nelson, S. G. Toxic Gas Sorbent and Process for Making Same. U.S. Patent 4,721,582, 1987. (17) Grim, R. E. Clay Mineralogy; McGraw-Hill Book Co.: New York, 1953. (18) Greenwood, N. N.; Earnshaw, A. Chemistry of the Elements; Pergamon Press: Oxford, U.K., 1984; pp 273-278. (19) Knozinger, H.; Ratnasamy, R. Catalytic Aluminas: Surface Models and Characterization of Surface Sites. In Catalysis Reviews: Science and Engineering; Marcel Dekker: New York, 1976; Vol. 17, No. 1, pp 31-70. (20) Murphy, J. F. Alumina in Air Pollution Control. Proceedings of the 102nd AIME Annual Meeting, Chicago, IL, 1972. (21) Bienstock, D. J.; Field, J. H.; Katell, S.; Plants, K. D. Evaluation of Dry Processes for Removing Sulfur Dioxide from Power Plant Flue Gases. JAPCA 1965, 15 (10), 459. (22) Schlesinger, M. D.; Illig, E. G. The Regeneration of Alkalized Alumina, Attritioning and SO2 Sorption Rates; U.S.B.M. Report No. 7275; U.S. Department of Interior, Bureau of Mines: Washington, DC, 1969. (23) Ma, W. T.; Chang, A. M.; Haslbeck, J. L.; Neal, N. G. NOXSO SO2/NOx Flue Gas Treatment Process Adsorption Chemistry and Kinetics. In Novel Adsorbents and their Environmental Application; AIChE Symposium Series 309; AIChE: 1995; Vol. 91, p 18. (24) Ma, W. T.; Haslbeck, J. L.; Neal, L. G.; Yeh, J. T. Life Cycle of the NOXSO SO2 and NOx Flue Gas Treatment Process: Process Modeling. Sep. Technol. 1991, 1, 195. (25) Yeh, J. T.; Ma, W. T.; Pennline, H. W.; Haslbeck, J. L.; Joubert, J. I.; Gromicko, F. N. Integrated Testing of the NOXSO Process: Simultaneous Removal of SO2 and NOx from Flue Gas. Chem. Eng. Commun. 1992, 114, 65. (26) Haslbeck, J. L.; Neal, L. G.; Perng, C. P.; Wang, C. J. Evaluation of the NOXSO Combined NOx/SO2 Flue Gas Treatment Process; U.S. DOE Report No. DOE/PC/7325-T2; Department of Energy: Washington, DC, 1985. (27) Haslbeck, J. L.; Wang, C. J.; Tseng, H. P.; Tucker, J. D. Evaluation of the NOXSO Combined NOx/SO2 Flue Gas Treatment Process; U.S. DOE Report No. DOE/FE/60148-T5; Department of Energy: Washington, DC, 1984.

Ind. Eng. Chem. Res., Vol. 37, No. 8, 1998 3381 (28) Gilbert, R. E.; Cox, D. F.; Hoflund, G. B. Computerinterfaced Digital Pulse Counting Circuit. Rev. Sci. Instrum. 1982, 53, 447. (29) Savistzky, A.; Golay, M. J. E. Smoothing and Differentiation of Data by Simplified Least Squares Procedures. Anal. Chem. 1964, 36, 1627. (30) Boehm, H. P. Acidic and Basic Properties of Hydroxylated Metal Oxide Surfaces. Discuss. Faraday Soc. 1972, 52, 264. (31) Sircar, S.; Rao, M. B.; Golden, T. C. Drying of Gases and Liquids by Activated Alumina. In Adsorption on New and Modified Inorganic Sorbents. Studies in Surface Science and Catalysis; Dabrowski, A., Tertykh, V. A., Eds.; Elsevier Science B.V.: New York, 1996; Vol. 99, p 629. (32) Coster, D. J.; Fripiat, J. J. Effect of Bulk Properties on the Rehydration of Aluminas. Langmuir 1995, 11, 2615. (33) Datta, A. Evidence for Cluster Sites on Catalytic Alumina. J. Phys. Chem. 1989, 93, 7053. (34) Knozinger, H. Specific Poisoning and Characterization of Catalytic Active Oxide Surfaces. In Advances in Catalysis; Academic Press: New York, 1976; Vol. 25, pp 184-267. (35) Parkyns, N. D. Adsorption Sites on Oxides, Infrared Studies of Adsorption of Oxides of Nitrogen. Proceedings of the Fifth International Congress on Catalysis, Miami Beach, FL, 1972.

(36) Lunsford, J. H. A Study of Surface Interactions on γ-alumina, Silica-Alumina, and Silica-Magnesia using the EPR Spectra of Adsorbed Nitric Oxide. J. Catal. 1969, 14, 379. (37) Terenin, A. N.; Roev, L. Infrared Spectra of Nitric Oxide Adsorbed on Transition Metals, their Salts and Oxides. Spectrochim. Acta 1959, 15, 946. (38) Jimenez-Gonzalez, A.; Schmeisser, D. Preparation and Spectroscopic Characterization of γ-Al2O3 Thin Films. Surf. Sci. 1991, 250, 59. (39) Krizek, J. Determination of Nitrogen Oxides in Small Concentrations. Chem. Prum. 1966, 16 (9), 558. (40) Hertzendorf, M. S. Air Pollution Control: A Guidebook to U.S. Regulations; Technomic Publishing Inc.: New York, 1973. (41) Himi, Y.; Muramatsu, F. Sampling Method for Determination of Nitrogen Oxides in Flue Gas. Jpn. Anal. 1969, 18 (6), 710.

Received for review February 18, 1998 Revised manuscript received May 24, 1998 Accepted May 25, 1998 IE9801047